Search

With billions of people facing moderate or severe food insecurity, the resilience of the global food supply will be of increasing concern due to the effects of climate change and geopolitical events. In this paper we describe a framework to better identify food security hotspots using a combination of remote sensing, deep learning, crop yield modeling, and causal modeling of the food distribution system. While we feel that the methods are adaptable to other regions of the world, we focus our analysis on the wheat breadbasket of northern India, which supplies a large percentage of the world's population. We present a quantitative analysis of deep learning domain adaptation methods for wheat farm identification based on curated remote sensing data from France. We model climate change impacts on crop yields using the existing crop yield modeling tool WOFOST and we identify key drivers of crop simulation error using a longitudinal penalized functional regression. A description of a system dynamics model of the food distribution system in India is also presented, along with results of food insecurity identification based on seeding this model with the predicted crop yields.

When learning an input-output mapping from very few examples, is it better to first infer a latent function that explains the examples, or is it better to directly predict new test outputs, e.g. using a neural network? We study this question on ARC, a highly diverse dataset of abstract reasoning tasks. We train neural models for induction (inferring latent functions) and transduction (directly predicting the test output for a given test input). Our models are trained on synthetic data generated by prompting LLMs to produce Python code specifying a function to be inferred, plus a stochastic subroutine for generating inputs to that function. We find inductive and transductive models solve very different problems, despite training on the same problems, and despite sharing the same neural architecture.

The growing use of Artificial Intelligence solutions has led to an explosion in image capture and its application in machine learning models. However, the lack of standardization in image quality generates inconsistencies in the results of these models. To mitigate this problem, Optical Character Recognition (OCR) is often used as a preprocessing technique, but it still faces challenges in scenarios with inadequate lighting, low resolution, and perspective distortions. This work aims to explore and evaluate various preprocessing techniques, such as grayscale conversion, CLAHE in RGB, and Bilateral Filter, applied to vehicle license plate recognition. Each technique is analyzed individually and in combination, using metrics such as accuracy, precision, recall, F1-score, ROC curve, AUC, and ANOVA, to identify the most effective method. The study uses a dataset of Brazilian vehicle license plates, widely used in OCR applications. The research provides a detailed analysis of best preprocessing practices, offering insights to optimize OCR performance in real-world scenarios., Comment: 12 pages, 13 figures

Collectible card games are challenging, widely played games that have received increasing attention from the AI research community in recent years. Despite important breakthroughs, the field still poses many unresolved challenges. This work aims to help further research on the genre by proposing a taxonomy of collectible card games by analyzing their rules, mechanics, and game modes from the perspective of game-playing AI research. To achieve this, we studied a set of popular games and provided a thorough discussion about their characteristics., Comment: 16 pages, accepted at the International Conference on Entertainment Computing (ICEC) 2024

We calculate the screening masses of pseudoscalar mesons in a hot and strongly magnetized medium within the framework of the SU(3) Nambu-Jona--Lasinio model, using a magnetic field-independent regularization scheme. Inverse magnetic catalysis (IMC) is implemented through the use of a magnetic field-dependent coupling $G(B)$, fitted to reproduce lattice QCD results for the pseudocritical chiral transition temperature $T_c^B$. For the external homogeneous magnetic field considered, neutral screening masses separate in two types: perpendicular and parallel to the direction of the field, while for charged mesons only parallel energies can be defined for each Landau level. We obtain $m_{\mathrm{scr},\perp} > m_{\mathrm{scr},\parallel}$, as expected from causality. Thermally, all screening energies are almost constant until some critical temperature, whose behavior is correlated with $T_c^B$. They rapidly increase around this value, keeping a steady enhancement afterwards due to thermal excitation. Magnetically, neutral parallel masses are enhanced (suppressed) at high temperatures when considering $G(B)$ ($G$). Perpendicular ones display a non-linear magnetic behavior for $G$ (due to increasing $T_c^B$), while they always increase with $B$ for $G(B)$. Charged parallel energies are also always magnetically enhanced. In the high-temperature limit, we show that both neutral and charged screening energies converge to $2\pi T$. At $B=0$ the model overestimates the remaining quark interaction in this regime. At $B\neq 0$ we find that, when IMC is accounted for, the interaction is suppressed as $B$ increases, a fact that appears to be at odds with currently available lattice QCD results., Comment: 38 pages, 11 figures

Major Depressive Disorder and anxiety disorders affect millions globally, contributing significantly to the burden of mental health issues. Early screening is crucial for effective intervention, as timely identification of mental health issues can significantly improve treatment outcomes. Artificial intelligence (AI) can be valuable for improving the screening of mental disorders, enabling early intervention and better treatment outcomes. AI-driven screening can leverage the analysis of multiple data sources, including facial features in digital images. However, existing methods often rely on controlled environments or specialized equipment, limiting their broad applicability. This study explores the potential of AI models for ubiquitous depression-anxiety screening given face-centric selfies. The investigation focuses on high-risk pregnant patients, a population that is particularly vulnerable to mental health issues. To cope with limited training data resulting from our clinical setup, pre-trained models were utilized in two different approaches: fine-tuning convolutional neural networks (CNNs) originally designed for facial expression recognition and employing vision-language models (VLMs) for zero-shot analysis of facial expressions. Experimental results indicate that the proposed VLM-based method significantly outperforms CNNs, achieving an accuracy of 77.6% and an F1-score of 56.0%. Although there is significant room for improvement, the results suggest that VLMs can be a promising approach for mental health screening, especially in scenarios with limited data.

The Problem-oriented AutoML in Clustering (PoAC) framework introduces a novel, flexible approach to automating clustering tasks by addressing the shortcomings of traditional AutoML solutions. Conventional methods often rely on predefined internal Clustering Validity Indexes (CVIs) and static meta-features, limiting their adaptability and effectiveness across diverse clustering tasks. In contrast, PoAC establishes a dynamic connection between the clustering problem, CVIs, and meta-features, allowing users to customize these components based on the specific context and goals of their task. At its core, PoAC employs a surrogate model trained on a large meta-knowledge base of previous clustering datasets and solutions, enabling it to infer the quality of new clustering pipelines and synthesize optimal solutions for unseen datasets. Unlike many AutoML frameworks that are constrained by fixed evaluation metrics and algorithm sets, PoAC is algorithm-agnostic, adapting seamlessly to different clustering problems without requiring additional data or retraining. Experimental results demonstrate that PoAC not only outperforms state-of-the-art frameworks on a variety of datasets but also excels in specific tasks such as data visualization, and highlight its ability to dynamically adjust pipeline configurations based on dataset complexity.

This paper is devoted to a complete characterization of the free boundary of a particular solution to the following spectral $k$-partition problem with measure and inclusion constraints: \[ \inf \left\{\sum_{i=1}^k \lambda_1(\omega_i)\; : \; \omega_i \subset \Omega \mbox{ are nonempty open sets for all } i=1,\ldots, k,\; \omega_i \cap \omega_j = \emptyset \: \text{for all}\: i \not=j \mbox{ and } \sum_{i=1}^{k}|\omega_i| = a \right\}, \] where $\Omega$ is a bounded domain of $\mathbb{R}^N$, $a\in (0,|\Omega|)$. In particular, we prove free boundary conditions, classify contact points, characterize the regular and singular part of the free boundary (including branching points), and describe the interaction of the partition with the fixed boundary $\partial \Omega$. The proof is based on a perturbed version of the problem, combined with monotonicity formulas, blowup analysis and classification of blowups, suitable deformations of optimal sets and eigenfunctions, as well as the improvement of flatness of [Russ-Trey-Velichkov, CVPDE 58, 2019] for the one-phase points, and of [De Philippis-Spolaor-Velichkov, Invent. Math. 225, 2021] at two-phase points., Comment: 39 pages

This paper demonstrates that a single-layer neural network using Parametric Rectified Linear Unit (PReLU) activation can solve the XOR problem, a simple fact that has been overlooked so far. We compare this solution to the multi-layer perceptron (MLP) and the Growing Cosine Unit (GCU) activation function and explain why PReLU enables this capability. Our results show that the single-layer PReLU network can achieve 100\% success rate in a wider range of learning rates while using only three learnable parameters.

We study the symmetry breaking and restoration behavior of a self-interacting charged scalar field theory under the influence of a constant electric field and finite temperature. Our study is performed in the context of the optimized perturbation theory. The dependence of the effective potential with constant electric fields is established by means of the bosonic propagators in the Schwinger proper-time method. Explicit analytical expressions for the electric and thermal contributions are found. Our results show a decreasing behavior for the vacuum expectation value as a function of the electric fields, which is strengthened by the temperature effect. A first-order phase transition that occurs at zero /weak electric fields changes to a second-order phase transition under strong electric fields. The critical temperature for the phase transition exhibited a very weak dependence on the electric field. Additionally, we computed the vacuum persistence probability rate for the interacting theory, finding a peak at the critical point. The maximum value of this rate at the critical point is found to be independent of the coupling constant but depended solely on the magnitude of the electric field.

We consider the interaction-round-a-face version of the isotropic six-vertex model. The associated spin chain is made of two coupled Heisenberg spin chains with different boundary twists. The phase diagram of the model and the long distance correlations were studied in [Nucl. Phys. B, 995 (2023) 116333]. Here, we compute the short-distance correlation functions of the model in the ground state for finite system sizes via non-linear integral equations and in the thermodynamic limit. This was possible since the model satisfies the face version of the discrete quantum Knizhnik-Zamolodchikov (qKZ) equation. A suitable ansatz for the density matrix is proposed in the form of a direct sum of two Heisenberg density matrices, which allows us to obtain the discrete functional equation for the two-site function $\omega(\lambda_1,\lambda_2)$. Thanks to the known results on the factorization of correlation functions of the Heisenberg chain, we are able to compute the density matrix of the IRF model for up to four sites and its associated spin chain for up to three sites., Comment: 25 pages, 4 figures, 1 table

This study presents a comprehensive methodology for modeling and forecasting the historical time series of active fire spots detected by the AQUA\_M-T satellite in the Amazon, Brazil. The approach employs a mixed Recurrent Neural Network (RNN) model, combining Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) architectures to predict the monthly accumulations of daily detected active fire spots. Data analysis revealed a consistent seasonality over time, with annual maximum and minimum values tending to repeat at the same periods each year. The primary objective is to verify whether the forecasts capture this inherent seasonality through machine learning techniques. The methodology involved careful data preparation, model configuration, and training using cross-validation with two seeds, ensuring that the data generalizes well to both the test and validation sets for both seeds. The results indicate that the combined LSTM and GRU model delivers excellent forecasting performance, demonstrating its effectiveness in capturing complex temporal patterns and modeling the observed time series. This research significantly contributes to the application of deep learning techniques in environmental monitoring, specifically in forecasting active fire spots. The proposed approach highlights the potential for adaptation to other time series forecasting challenges, opening new opportunities for research and development in machine learning and prediction of natural phenomena. Keywords: Time Series Forecasting; Recurrent Neural Networks; Deep Learning., Comment: 16 pages and 24 figures, in Portuguese language

Recent advances in audio-text cross-modal contrastive learning have shown its potential towards zero-shot learning. One possibility for this is by projecting item embeddings from pre-trained backbone neural networks into a cross-modal space in which item similarity can be calculated in either domain. This process relies on a strong unimodal pre-training of the backbone networks, and on a data-intensive training task for the projectors. These two processes can be biased by unintentional data leakage, which can arise from using supervised learning in pre-training or from inadvertently training the cross-modal projection using labels from the zero-shot learning evaluation. In this study, we show that a significant part of the measured zero-shot learning accuracy is due to strengths inherited from the audio and text backbones, that is, they are not learned in the cross-modal domain and are not transferred from one modality to another.

Topological defect lines (TDLs) in two-dimensional conformal field theories (CFTs) are standard examples of generalized symmetries in quantum field theory. Integrable lattice incarnations of these TDLs, such as those provided by spin/anyonic chains, provide a crucial playground to investigate their properties, both analytically and numerically. Here, a family of parameter-dependent integrable lattice models is presented, which realize different TDLs in a given CFT as the parameter is varied. These models are based on the general quantum-inverse scattering construction, and involve inhomogeneities of the spectral parameter. Both defect hamiltonians and (defect) line operators are obtained in closed form. By varying the inhomogeneities, renormalization group flows between different TDLs (such as the Verlinde lines associated with the Virasoro primaries $(1,s)$ and $(s,1)$ in diagonal minimal CFTs) are then studied using different aspects of the Bethe-ansatz as well as ab-initio numerical techniques. Relationships with the anisotropic Kondo model as well as its non-Hermitian version are briefly discussed, Comment: 39 pages, 10 figures

We introduce a mechanism by which a misaligned ALP can be dynamically converted into a dark photon in the presence of a background magnetic field. An abundance of non-relativistic ALPs will convert to dark photons with momentum of order the inhomogeneities in the background field; therefore a highly homogeneous field will produce non-relativistic dark photons without relying on any redshifting of their momenta. Taking hidden sector magnetic fields produced by a first order phase transition, the mechanism can reproduce the relic abundance of dark matter for a wide range of dark photon masses down to $10^{-13}$ eV., Comment: 23 pages, 8 figures

Let $f: M\rightarrow M$ be a continuous map on a compact metric space $M$ equipped with a fixed metric $d$, and let $\tau$ be the topology on $M$ induced by $d$. First, we will establish some fundamental properties of the mean Hausdorff dimension. Furthermore, it is important to note that the metric mean dimension and mean Hausdorff dimension depend on the metric chosen for $M$. In this work, we will prove that, for a fixed dynamical system $f:M\rightarrow M$, the functions $\text{mdim}_{\text{M}}(M, f):M(\tau)\rightarrow \mathbb{R}\cup \{\infty\}$ and $\text{mdim}_{\text{H}}(M, f):M(\tau)\rightarrow \mathbb{R}\cup \{\infty\}$ are not continuous. Here, $ \text{mdim}_{\text{M}}(M, f)(\rho)= \text{mdim}_{\text{M}}(M,\rho, f)$ and $ \text{mdim}_{\text{H}}(M, f)(\rho)= \text{mdim}_{\text{H}}(M,\rho, f)$ represent, respectively, the metric mean dimension and the mean Hausdorff dimension of $f$ with respect to $\rho\in M(\tau)$ and $M(\tau)$ is the set consisting of all equivalent metrics to $d$ on $M$. Furthermore, we will present examples of certain classes of metrics for which the metric mean dimension is a continuous function.

In settings where only a budgeted amount of labeled data can be afforded, active learning seeks to devise query strategies for selecting the most informative data points to be labeled, aiming to enhance learning algorithms' efficiency and performance. Numerous such query strategies have been proposed and compared in the active learning literature. However, the community still lacks standardized benchmarks for comparing the performance of different query strategies. This particularly holds for the combination of query strategies with different learning algorithms into active learning pipelines and examining the impact of the learning algorithm choice. To close this gap, we propose ALPBench, which facilitates the specification, execution, and performance monitoring of active learning pipelines. It has built-in measures to ensure evaluations are done reproducibly, saving exact dataset splits and hyperparameter settings of used algorithms. In total, ALPBench consists of 86 real-world tabular classification datasets and 5 active learning settings, yielding 430 active learning problems. To demonstrate its usefulness and broad compatibility with various learning algorithms and query strategies, we conduct an exemplary study evaluating 9 query strategies paired with 8 learning algorithms in 2 different settings. We provide ALPBench here: https://github.com/ValentinMargraf/ActiveLearningPipelines.

Dental panoramic radiographs offer vast diagnostic opportunities, but training supervised deep learning networks for automatic analysis of those radiology images is hampered by a shortage of labeled data. Here, a different perspective on this problem is introduced. A semi-supervised learning framework is proposed to classify thirteen dental conditions on panoramic radiographs, with a particular emphasis on teeth. Large language models were explored to annotate the most common dental conditions based on dental reports. Additionally, a masked autoencoder was employed to pre-train the classification neural network, and a Vision Transformer was used to leverage the unlabeled data. The analyses were validated using two of the most extensive datasets in the literature, comprising 8,795 panoramic radiographs and 8,029 paired reports and images. Encouragingly, the results consistently met or surpassed the baseline metrics for the Matthews correlation coefficient. A comparison of the proposed solution with human practitioners, supported by statistical analysis, highlighted its effectiveness and performance limitations; based on the degree of agreement among specialists, the solution demonstrated an accuracy level comparable to that of a junior specialist., Comment: 43 pages, 12 figures, 9 tables

We consider a multiphase spectral problem on a stratified Lie group. We prove the existence of an eigenfunction of $(2,q)$-eigenvalue problem on a bounded domain. Furthermore, we also establish a Pohozaev-like identity corresponding to the problem on the Heisenberg group.

Multiphoton indistinguishability is a central resource for quantum enhancement in sensing and computation. Developing and certifying large scale photonic devices requires reliable and accurate characterization of this resource, preferably using methods that are robust against experimental errors. Here, we propose a set of methods for the characterization of multiphoton indistinguishability, based on measurements of bunching and photon number variance. Our methods are robust in a semi-device independent way, in the sense of being effective even when the interferometers are incorrectly dialled. We demonstrate the effectiveness of this approach using an advanced photonic platform comprising a quantum-dot single-photon source and a universal fully-programmable integrated photonic processor. Our results show the practical usefulness of our methods, providing robust certification tools that can be scaled up to larger systems., Comment: 19 pages + 10 figures

This paper studies Schauder theory to transmission problems modelled by fully nonlinear uniformly elliptic equations of second order. We focus on operators F that fails to be concave or convex in the space of symmetric matrices. In a first scenario, it is considered that F enjoys a small ellipticity aperture. In our second case, we study regularity results where the convexity of the superlevel (or sublevel) sets is verified, implying that the operator F is quasiconcave (or quasiconvex).

Depression has affected millions of people worldwide and has become one of the most common mental disorders. Early mental disorder detection can reduce costs for public health agencies and prevent other major comorbidities. Additionally, the shortage of specialized personnel is very concerning since Depression diagnosis is highly dependent on expert professionals and is time-consuming. Recent research has evidenced that machine learning (ML) and Natural Language Processing (NLP) tools and techniques have significantly bene ted the diagnosis of depression. However, there are still several challenges in the assessment of depression detection approaches in which other conditions such as post-traumatic stress disorder (PTSD) are present. These challenges include assessing alternatives in terms of data cleaning and pre-processing techniques, feature selection, and appropriate ML classification algorithms. This paper tackels such an assessment based on a case study that compares different ML classifiers, specifically in terms of data cleaning and pre-processing, feature selection, parameter setting, and model choices. The case study is based on the Distress Analysis Interview Corpus - Wizard-of-Oz (DAIC-WOZ) dataset, which is designed to support the diagnosis of mental disorders such as depression, anxiety, and PTSD. Besides the assessment of alternative techniques, we were able to build models with accuracy levels around 84% with Random Forest and XGBoost models, which is significantly higher than the results from the comparable literature which presented the level of accuracy of 72% from the SVM model.

Understanding the decisions of tree-based ensembles and their relationships is pivotal for machine learning model interpretation. Recent attempts to mitigate the human-in-the-loop interpretation challenge have explored the extraction of the decision structure underlying the model taking advantage of graph simplification and path emphasis. However, while these efforts enhance the visualisation experience, they may either result in a visually complex representation or compromise the interpretability of the original ensemble model. In addressing this challenge, especially in complex scenarios, we introduce the Decision Predicate Graph (DPG) as a model-agnostic tool to provide a global interpretation of the model. DPG is a graph structure that captures the tree-based ensemble model and learned dataset details, preserving the relations among features, logical decisions, and predictions towards emphasising insightful points. Leveraging well-known graph theory concepts, such as the notions of centrality and community, DPG offers additional quantitative insights into the model, complementing visualisation techniques, expanding the problem space descriptions, and offering diverse possibilities for extensions. Empirical experiments demonstrate the potential of DPG in addressing traditional benchmarks and complex classification scenarios.